Migraine Cortical Spreading Depression Model

ISEF Category: Computational Biology and Bioinformatics

Subcategory: Computational Neuroscience  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

A migraine aura can spread across the brain like a slow wave of electric silence. That wave has a name, cortical spreading depression. If you can model that spread on a cortical mesh, you can test how magnesium or CGRP-related changes might slow it down. That turns a painful human problem into a clean simulation question.

What Is It?

This project models cortical spreading depression, or CSD, which is a moving wave of reduced brain activity that can help explain migraine aura. Think of it like a line of falling dominoes across the cortex. One patch of tissue becomes unstable, then the effect spreads to nearby tissue.

You would build that cortex as a mesh, which is a digital surface made of many small triangles. FreeSurfer can help turn brain imaging data into that surface. Brian2 can then simulate how activity changes across the mesh when you change parameters tied to magnesium or CGRP receptors, a migraine-linked signaling system. Your job is not to prove a medicine works in people. Your job is to test how the model responds when you perturb key inputs.

Why This Is a Good Topic

This is a strong science fair topic because the main question is clear, measurable, and simulation-based. You can change one variable, run the model again, and compare wave speed, spread distance, or activity decay. The project connects to migraine research, drug screening ideas, and brain dynamics, but you can study it without a wet lab. You will also learn how scientists turn real anatomy into a computational model and how they test whether a model behaves in a believable way.

Research Questions

• How does lowering the simulated magnesium level change the speed of cortical spreading depression across the mesh?
• What is the effect of CGRP-receptor perturbation on the total area reached by the wave?
• Does changing the mesh resolution alter the measured spread speed or wave shape?
• To what extent do combined magnesium and CGRP changes interact in the model, compared with each change alone?
• Which cortical regions on the mesh show the earliest and latest simulated spread?
• How does the chosen excitability threshold change whether the wave dies out or keeps propagating?

Basic Materials

• Computer with enough RAM to run Python simulations.
• Python installed with scientific libraries.
• Brian2 for neural and differential equation simulations.
• FreeSurfer for cortical surface processing.
• Git or another version control tool.
• Basic spreadsheet software for tracking runs and parameters.
• Headphones or speakers only if you plan to listen to alert sounds during long runs, not for the science itself.
• Notebook for recording parameter sets and output metrics.

Advanced Materials

• University or lab workstation with high-RAM CPU access.
• High-quality structural MRI data, if you plan to build a subject-specific mesh.
• FreeSurfer output files for cortical reconstruction and surface extraction.
• Brian2 plus any custom code for spatially extended excitable media.
• Python packages for mesh handling, such as nibabel, numpy, scipy, and matplotlib.
• A job scheduler or batch system for running large parameter sweeps.
• Statistical analysis software for mixed models or permutation tests.
• Optional GPU resources if you test a faster surrogate model.

Software & Tools

• Python: Runs the simulation code, processes outputs, and handles parameter sweeps.
• Brian2: Simulates the biophysical dynamics that drive wave spread across the cortex.
• FreeSurfer: Converts MRI data into a cortical surface mesh for the model.
• ImageJ: Helps inspect exported heatmaps or frame sequences from simulation output.
• Jupyter Notebook: Keeps code, plots, and notes together for fast iteration.

Experiment Steps

1. Define the exact output you will measure, such as wave speed, spread area, or time to extinction.
2. Choose the simplest cortical mesh that still preserves the anatomy you need for the question.
3. Build a baseline model and confirm that the wave behaves like cortical spreading depression, not random noise.
4. Plan one variable at a time for magnesium, CGRP-receptor perturbation, or mesh settings.
5. Design controls that separate real parameter effects from artifacts caused by mesh resolution or solver settings.
6. Decide how you will compare repeated runs, then preselect the statistics you will use before you start sweeping parameters.

Common Pitfalls

• Using a mesh that is too coarse, which hides local spread patterns and makes the wave look artificially smooth.
• Changing multiple parameters at once, which makes it impossible to tell whether magnesium or CGRP caused the effect.
• Treating the simulation as proof of a drug effect in people, which overstates what the model can support.
• Ignoring solver settings or time-step choice, which can create fake differences between runs.
• Measuring only one output, which can miss cases where the wave slows down but still covers the same total area.

What Makes This Competitive

A class project can stop at one simulation and one plot. A stronger project tests several parameter ranges, checks sensitivity to mesh choice, and compares at least two readouts, such as speed and spread area. You can also make the project stronger by validating the baseline model against published CSD behavior before testing magnesium or CGRP perturbations. Careful statistics and clear limits on what the model can and cannot claim will help a lot.

Project Variations

• Test how the same migraine model behaves on subject-specific meshes versus a generic cortical surface.
• Compare magnesium perturbation effects with CGRP-receptor perturbation effects across several excitability settings.
• Add a spatial lesion or region-of-interest trigger and measure how starting location changes wave propagation.

Learn More

• PubMed: Search for review articles on cortical spreading depression, migraine aura, magnesium, and CGRP signaling.
• NIH PubMed Central: Read free full-text neuroscience and migraine papers, especially methods sections with modeling details.
• MIT OpenCourseWare: Find free courses on computational neuroscience and dynamical systems to strengthen your model setup.
• FreeSurfer Documentation: Learn how cortical surfaces are reconstructed from MRI data and exported for analysis.
• Brian2 Documentation: Study examples for spiking and differential equation models, then adapt the framework to excitable tissue.
• NIH NCBI Bookshelf: Look up background chapters on migraine biology, brain excitability, and neural signaling.